Accumulation of Phase-Shift Nanoemulsions to Enhance MR-Guided Ultrasound-Mediated Tumor Ablation In Vivo

Magnetic resonance-guided high intensity focused ultrasound (MRgHIFU) is being explored as a non-invasive technology to treat solid tumors. However, the clinical use of HIFU for tumor ablation applications is currently limited by the long treatment times required. Phase-shift nanoemulsions (PSNE), consisting of liquid perfluorocarbon droplets that can be vaporized into microbubbles, are being developed to accelerate HIFU-mediated heating. The purpose of this study was to examine accumulation of PSNE in intramuscular rabbit tumors in vivo. MR images were acquired before and after intravenous injection of gadolinium-containing PSNE. MR signal enhancement was observed in rabbit tumors up to six hours after injection, indicating that PSNE accumulated in the tumors. In addition, PSNE vaporization was detected in the tumor with B-mode ultrasound imaging, and MR thermometry measurements indicated that PSNE accelerated the rate of HIFU-mediated heating. These results suggest that PSNE could dramatically improve the efficiency and clinical feasibility of MRgHIFU.


INTRODUCTION
Cancer is the second leading cause of death in the United States.More than 1 in 3 Americans develop cancer during their lifetime [1,2].Current standard-of-care includes surgery, chemotherapy, or radiotherapy.Surgery is generally preferred for localized tumors due to the serious systemic side effects often induced by chemotherapy and radiotherapy.However, surgery is limited to resectable tumors and many advanced tumors are inoperable.Several minimally invasive treatments have been developed for non-resectable solid tumors, which kill cancer cells by rapidly heating or freezing the tumor using microwaves, radiofrequency waves, lasers, or cryogens such as liquid nitrogen [3].However, these techniques generally require interstitial energy sources, which can cause complications from visceral and vascular puncture or tumor reseeding [4,5].To overcome these challenges, magnetic resonance-guided high intensity focused ultrasound (MRgHIFU) has been explored as a non-invasive technology to treat solid tumors.
High intensity focused ultrasound (HIFU) can be used to heat tissues rapidly and locally, forming lesions of coagulated tissue with millimeter precision [6][7][8][9][10][11][12].Ultrasound waves can pass through the skin and soft tissue without causing damage outside of the focal volume.HIFU is currently FDA-approved for treatment of uterine fibroids and is also being investigated clinically to thermally ablate tumors of the breast [13,14], pancreas [15], liver [16,17], kidney [16,18], and prostate [19][20][21].Non-invasive imaging is used to detect the location and boundary of the tumor and guide the placement of the ultrasound focus within the tumor.Magnetic resonance imaging (MRI) has been used extensively to guide HIFU treatments due to its high spatial resolution and contrast [22].In addition, MRI-based thermometry can be used to measure temperature variations during the HIFU treatment [23][24][25][26][27][28][29][30].Thus, MRI is useful for treatment planning and monitoring of HIFU treatment.
The clinical use of HIFU for tumor ablation applications is currently limited by the long treatment times (on the order of hours) and high acoustic powers required.Lesions formed by HIFU ablation are typically less than 10 mm 3 ; thus, multiple lesions must be formed sequentially to ablate the entire tumor volume.In order for HIFU ablation therapy to become clinically feasible, the treatment times must decrease by increasing the heating rate and/or lesion volumes achieved with each ultrasound exposure.In addition, adverse events including skin burns have been reported as a result of the high power required for consistent lesion formation [31,32].Thus, reducing the acoustic power needed for lesion formation would also improve the clinical utility of HIFU ablation therapy.
It has been shown that microbubbles can be used to accelerate ultrasound-mediated heating and lesion formation [33][34][35][36][37][38][39].At sufficiently high acoustic pressures, ultrasound can cause bubbles to oscillate non-linearly followed by a violent collapse known as inertial cavitation, resulting in acoustic emissions across a broad range of frequencies.Tissue absorption of acoustic energy is converted into heat, which increases with acoustic frequency.Thus, the high-frequency acoustic emissions from inertial cavitation accelerate heating within tissue.Studies in tissue-mimicking gel phantoms [34], and in ex vivo [33], and in vivo tissue [35] have demonstrated a dramatic increase in the measured heating rate and peak temperature in the presence of acoustic cavitation.These results indicate that microbubbles can be used to reduce the time required for thermal ablation.
Because bubbles are not readily available in tumors, they must be formed or introduced into the tissue.A common approach for introducing microbubbles is to systemically 110 Accumulation of Phase-Shift Nanoemulsions to Enhance MR-Guided Ultrasound-Mediated Tumor Ablation In Vivo administer ultrasound contrast agents (UCA), which are microbubbles coated with a lipid, protein, or polymer shell, before ultrasound exposure [37,38,[40][41][42].However, microbubbles injected into the bloodstream remain in circulation and do not accumulate within the tumor due to their size [43][44][45].The circulating bubbles can increase the absorption of ultrasound in tissue along the path of the propagating waves, potentially causing unwanted heating and irreversible thermal damage to healthy tissue surrounding the targeted tumor.Alternatively, bubbles can be formed locally within the tumor using very high amplitude acoustic pulses [46][47][48][49].However, the production of bubbles in tumors using ultrasound pulses can vary significantly due to the heterogeneity of tumor tissue, making it difficult to reliably control bubble formation.Furthermore, it has been documented that the pressure threshold for bubble formation in tissue without nuclei can exceed 10 MPa [48,49], and controlling bubble formation temporally and subsequent oscillations at these pressures is extremely difficult.Phase-shift nanoemulsions (PSNE) present a potential solution to these challenges.PSNE are suspensions of lipid-or albumin-coated liquid perfluorocarbon droplets (Figure 1) which exist in a superheated state, such that the surface tension of the droplet shell prevents vaporization of the liquid perfluorocarbon core even at temperatures above the bulk perfluorocarbon boiling point.
Ultrasound pulses can be used to induce a phase change in the emulsion droplets in response to the rarefactional part of the pressure wave, converting the liquid droplets into gas-filled microbubbles [50][51][52].
In general, blood vessels in tumors are inherently leakier than normal vessels.Thus, nanoparticles can preferentially accumulate in tumors by passing through fenestrations in the tumor vasculature, a phenomenon known as the enhanced permeability and retention (EPR) effect [53,54].In contrast to exogenous microbubbles that are Illustration of a perfluorocarbon droplet in PSNE.A droplet is composed of a liquid perfluorocarbon core surrounded by a phospholipid shell containing polyethylene glycol (PEG) to improve stability in vivo.
generally a few microns in diameter, liquid droplets can be produced with diameters of 200 nm or less in order to achieve passive accumulation in tumors through the EPR effect [55,56].After accumulation of PSNE in the tumor, short ultrasound pulses of sufficient pressures can be used to vaporize the droplets, forming perfluorocarbon gas bubbles in situ that can nucleate inertial cavitation [55].The bubbles are only formed at the transducer focus, thus localizing the bubble-enhanced heating and ablation to the transducer focal volume.
In this in vivo study, accumulation of PSNE in rabbit VX2 tumors was determined as a function of time after injection using T 1 -weighted MRI.PSNE was tagged with gadolinium-chelate (Gd), an MR contrast agent, in order to detect it through a change in MR signal intensity.In addition, ultrasound-triggered vaporization of PSNE was tested and an ultrasound imaging system was used to detect the resulting formation of microbubbles.Furthermore, MR thermometry measurements were acquired in order to determine the amount of heating in the tumor with and without PSNE.The overall aim was to investigate the accumulation and activity of PSNE in rabbit VX2 tumors in vivo, in order to assess the potential for PSNE to improve the efficiency and clinical feasibility of MR-guided HIFU for non-invasive treatment of non-resectable solid tumors.

METHODS 2.1. Preparation of Phase-Shift Nanoemulsions
Lipids were obtained from Avanti Lipids (Alabaster, AL, USA), gadoliniumdiethylenetriaminepentaacetate-bis-oleate (Gd-DTPA-BOA) was purchased from Chemir (Maryland Heights, MO, USA), and the liquid perfluorocarbon was obtained from Fluoromed (Round Rock, TX, USA).Phase-shift nanoemulsions containing gadolinium chelate (Gd-PSNE) were composed of 1,2-dipalmitoyl-sn-glycero-3phosphocholine (DPPC), 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[amino(polyethylene glycol)-2000] (DSPE-PEG2000), and Gd-DTPA-BOA in a molar ratio of 64:6:30 (DPPC:DSPE-PEG2000:Gd-DTPA-BOA).For HIFU experiments, standard PSNE without Gd-DTPA-BOA were prepared using the same procedure with a molar ratio of 94:6 DPPC:DSPE-PEG2000.Lipids (190 mg) were dissolved in chloroform in a glass round-bottom flask and the solvent was evaporated to form a dry lipid film.After dessicating overnight, the lipid film was rehydrated with 10 ml phosphate-buffered saline (PBS) at 45°C and sonicated for 2 minutes at 20% power using a Vibra-Cell sonicator probe (Sonics & Materials, Newtown, CT, USA) to disperse the lipids.Liquid perfluorocarbon dodecafluoropentane (DDFP, C 5 F 12 ) was added to the solution at a volume ratio of 4% and the sample was sonicated in an ice-water bath using the Vibra-Cell sonicator at 25% power in a pulsed sequence of 10 s on, 50 s off, for a total insonation time of 60 s.After sonication, the solution was extruded 16 times through a LIPEX extruder (Northern Lipids, Burnaby, BC, Canada) with polycarbonate membrane filters containing 200 nm pores (Whatman, Piscataway, NJ, USA).The resulting nanoemulsion was dialyzed for 24 hours in 1 L of PBS using 50 kDa dialysis membranes (Spectrum Labs, Rancho Dominguez, CA, USA) in order to remove any gadolinium chelate that was not incorporated into the perfluorocarbon 112 Accumulation of Phase-Shift Nanoemulsions to Enhance MR-Guided Ultrasound-Mediated Tumor Ablation In Vivo droplets.The gadolinium concentration in the Gd-PSNE solution was 12 mM before dialysis.PSNE solutions were stored at 4°C until use.Size distribution measurements were acquired at 37°C using dynamic light scattering (90Plus Particle Size Analyzer, Brookhaven Instruments, Holtsville, NY, USA).

Animals
All animal experiments were performed according to protocols approved by the Harvard Medical School Institutional Animal Care and Use Committee.Two weeks before each experiment, 3 × 10 7 VX2 cells were injected directly into the left thigh of 3 kg New Zealand white rabbits at a depth of approximately 1 cm.During each experiment, animals were anesthetized with a mixture of ketamine (40 mg/kg body weight, Aveco, Fort Dodge, IA, USA) and xylazine (10 mg/kg body weight, Lloyd Laboratories, Shenandoah, IA, USA).PSNE (0.5 ml/kg body weight) was injected intravenously through a catheter in the ear vein and flushed with 0.5 ml saline.Nine animals were used in this study: four to assess Gd-PSNE accumulation in tumors, one to investigate PSNE vaporization in the tumor, and four to measure HIFU-mediated heating in tumors with and without PSNE.

Experimental Setup
A diagram of the experimental setup is shown in Figure 2.For all MR imaging experiments, each animal was placed on a custom-built platform in a clinical GE Signa 3T short-bore MRI scanner, oriented so that the tumor was located at the center of a custom-built circular surface MR coil.A custom-built, MR-compatible HIFU transducer (1.5-MHz center frequency, 10-cm aperture, and 8-cm radius of curvature) was placed in a water tank located below the coil.The animal was oriented so that the surface of the  Setup for HIFU experiments.
tumor-bearing thigh was immersed in the water to ensure adequate acoustic coupling for ultrasound exposures.T 1 -weighted MR images were acquired in all three planes at six time points: before injection of Gd-PSNE and at 0.5, 1, 2, 4, and 6 hours after injection.In each plane, 11-18 images were acquired using a fast spin-echo MR sequence with a pulse repetition time (TR) of 500 ms and echo time (TE) of 13.2 ms.Square 12-cm slices were acquired with a slice thickness of 2 mm and a resolution of 256 × 256 pixels.MR thermometry measurements were acquired using a phase-difference fast spoiled gradient-echo MR sequence with TR of 27 ms and TE of 13.5 ms.Square 8-cm slices were acquired with a slice thickness of 2 mm and a resolution of 256 × 256 pixels.A total of forty temperature maps were acquired at intervals of 3.4 s.To detect changes in the tumor due to thermal damage, T 2 -weighted images were acquired using a fast spin-echo sequence with a TR of 4500 ms and a TE of 84.9 ms.Square 12-cm slices were acquired with a slice thickness of 2 mm and a resolution of 256 × 256 pixels.
The HIFU transducer was driven by waveforms produced by a function generator and amplified with an RF power amplifier (240L, E&I, Rochester, NY) that was connected to the HIFU transducer through a BNC connector panel in the wall of the MR control room.Two function generators were used in tandem to generate the ultrasound signals for HIFU ablation experiments.For PSNE vaporization, 100-cycle, 90 W pulses were generated with one arbitrary waveform generator (33220A, Agilent, Santa Clara, CA, USA) at a pulse repetition frequency (PRF) of 670 ms, immediately followed by 1 million-cycle signals at lower acoustic powers (between 5.2 W and 38.5 W) generated by the second arbitrary waveform generator (395, Wavetek, San Diego, CA, USA) to drive inertial cavitation and heat the tumor.The sonication duration was set to either 10 s or 30 s, depending on the acoustic power level.The acoustic output was calibrated using a force balance method [57].At the beginning of the experiment, the location of the HIFU focus relative to the MRI coordinates was determined.An agar gel phantom was heated with a 10 W, continuous-wave ultrasound signal for 30 s and an MR thermometry scan was acquired in order to identify the location of peak temperature increase, which was assumed to be the focal point.In a separate experiment without MRI, a 10L5 imaging array on a portable ultrasound system (Terason 2000, Terason, Burlington, MA, USA) was used to acquire B-mode images of PSNE vaporization.All HIFU experiments were performed two hours after PSNE injection.

MR Data Analysis
To evaluate Gd-PSNE accumulation in rabbit VX2 tumors over time, T 1 -weighted MR images were post-processed in MATLAB (Mathworks, Natick, MA, USA).Each image was cropped to isolate the tumor region, Gaussian-filtered, and an intensity threshold was applied.The cutoff value was selected based on the maximum signal intensity in a region of interest within the tumor before PSNE injection.The number of voxels that exceeded the threshold MR signal intensity was counted to determine the total enhanced volume.The dimensions of each voxel were 0.47 mm × 0.47 mm × 2.00 mm.
For MR-guided HIFU experiments, changes in temperature were determined from phase-difference maps as previously described [28].Thermal dose maps were computed from the temperature maps using Equation 1 [28]: Accumulation of Phase-Shift Nanoemulsions to Enhance MR-Guided Ultrasound-Mediated Tumor Ablation In Vivo (1) where CEM 43 represents the cumulative thermal dose in cumulative equivalent minutes at 43°C, R is a constant value of 0.5 for temperatures above 43°C or 0.25 for temperatures below 43°C, and T represents the mean temperature (°C) measured during time ∆t (minutes).

Accumulation of Gd-PSNE in Tumor
The size distributions of PSNE in this study were measured with dynamic light scattering and are shown in Figure 3.The mean size of Gd-PSNE was 189 nm and the mean size of standard PSNE (without Gd) was 158 nm.The zeta potential of PSNE was -5.5 ± 14.8 mV.Although standard PSNE could not be detected with MRI, it is expected that the accumulation of standard PSNE in tumors would be comparable to Gd-PSNE because of the similarity in size.In fact, it is possible that greater accumulation of standard PSNE could occur due to the smaller size compared to Gd-PSNE.To study the in vivo accumulation of Gd-PSNE in rabbit tumors over time, T 1 -weighted MR images were acquired before and after intravenous injection of Gd-PSNE.Representative images of a rabbit tumor with and without Gd-PSNE are shown in Figure 4. Signal enhancement is evident around the periphery of the tumor, which is expected to have a larger vascular bed and higher perfusion.In the poorly perfused necrotic core, however, significantly less accumulation was observed.This pattern of PSNE accumulation was observed in all MR image slices of tumors with necrotic cores.The total tumor volume with enhanced MR signal after Gd-PSNE injection was determined from T 1 -weighted images in four rabbits and is plotted in Figure 5.The total volume with enhanced MR signal increased rapidly in the first 30 minutes after injection of Gd-PSNE, followed by a slight increase over six hours after injection.Because very little change in accumulation was observed between two and six hours after injection, PSNE vaporization and MR-guided HIFU ablation experiments were conducted two hours after PSNE injection.

HIFU-Mediated Vaporization and Heating in Tumor
To investigate acoustic droplet vaporization (ADV) of PSNE in vivo, a 90-W, 67-µs HIFU pulse was transmitted into the rabbit tumor to induce ADV of PSNE, and B-mode ultrasound was used to detect bubble formation.Representative B-mode ultrasound images of the tumor before and after PSNE vaporization are shown in Figure 6.The  signal intensity increased at the HIFU focus after the ADV pulse, indicating the presence of bubbles that were formed due to PSNE vaporization.The enhancement is even more noticeable on the corresponding background-subtracted images.The bubbles dissolved into the tissue within a few seconds but that timeframe was sufficient to nucleate acoustic cavitation for bubble-enhanced heating.
To quantify the amount of HIFU-mediated heating in the tumor, MR thermometry measurements were acquired during sonication, and the temperature was calculated at multiple time points.The relative difference in temperature in tumors with PSNE compared to tumors without PSNE is shown in Figure 7 as a function of time during and after sonication using identical ultrasound parameters (acoustic power of 5.2 W).The peak temperature measured in the tumor with PSNE was 60°C, compared to 56°C without PSNE.While the difference in temperature seems minor, the impact on cells and tissue can be significant.For example, Dewhirst et al. [58] demonstrated that increasing the temperature from 55°C to 60°C ( +9%) reduced the heating time required for cellular necrosis from 30 sec to 5.1 sec (-83%).In our study, this increase in temperature due to PSNE vaporization was associated with a dramatic increase in the applied thermal dose, which is a better predictor of thermal damage.In the absence of vaporized PSNE, the thermal dose calculated from MR thermometry was 51 equivalent minutes at 43°C.When PSNE was vaporized and the peak temperature was increased to 60°C, the applied thermal dose was increased by a factor of 45 to 2296 equivalent minutes at 43°C.It is evident that for these acoustic parameters, the bubbles dramatically increased the efficiency with which ultrasound was absorbed and the tumor was heated.More importantly, the enhancement in ultrasound-mediated heating led to a reduction in the acoustic power required for lesion formation within the tumors.To illustrate this, thermal dose maps are shown in Figure 8

DISCUSSION
The results of this study demonstrate that Gd-PSNE can accumulate in rabbit tumors for at least six hours after intravenous injection.This suggests that the time window for conducting HIFU ablations after a single PSNE injection is large (on the order of hours).In contrast, microbubble-enhanced HIFU ablation requires continuous infusion or repeated injections due to lack of microbubble accumulation in tumors and rapid clearance from circulation (half-life < 5 min).MR signal enhancement (indicating the presence of Gd-PSNE) was most evident around the tumor rim and appeared to gradually diffuse further into the tumor over time, but enhancement was not observed at the center of the tumor.This is likely due to the fact that the center of the tumor is typically poorly perfused and necrotic in advanced VX2 tumors, which could pose a challenge to the delivery of therapeutic agents and nanoparticles [63][64][65].In contrast, the tumor rim is generally highly vascularized and contains the majority of the viable cancer cells, which would be more easily targeted through the EPR effect [66].More importantly, PSNE collected within highly vascularized regions could be vaporized to form microbubbles that dramatically enhanced HIFU-mediated heating, thus counteracting the cooling effects provided by circulating blood.Furthermore, HIFU-mediated ablation is problematic throughout well-perfused tissues, such as the periphery of tumors, due to cooling effects of circulating blood.Similar to other nanoparticles, PSNE accumulates predominately in the tumor periphery, seeding highly vascularized regions with nuclei for inertial cavitation, which can be used to enhance HIFU-mediated ablation as well as destroy vessels that feed tumors.Ultimately, this will reduce blood perfusion in solid tumors, potentially leading to an improvement in the efficiency of HIFU-induced heating in the tumor core.It is important to note that HIFU ablation treatments require constant monitoring and adjustment of the transmitted acoustic power and sonication duration in order to adequately ablate the targeted structures.This will also be necessary with PSNE, but inertial cavitation emissions from vaporized PSNE can potentially be monitored to serve as an additional feedback parameter [67][68][69].HIFU-mediated vaporization of PSNE was observed with B-mode ultrasound imaging of the tumor two hours after intravenous injection of PSNE.Previous studies have used microbubbles administered systemically in order to enhance HIFU-mediated heating.However, this approach is limited by the rapid clearance time (on the order of minutes) and the fact that the size of the microbubbles (microns) prevents extravasation through tumor vessels.As an alternative to microbubbles, Zhang et al. [70] studied lesion formation in canine liver using perfluorocarbon droplets with diameters greater than a micron.Most likely, those droplets were vaporized within the tumor vasculature, whereas the nanoemulsion droplets used in this study were small enough (diameter < 200 nm) to extravasate through leaky tumor vessel walls and accumulate within the tumor interstitium and reside for hours.This study was the first in vivo demonstration that PSNE could be vaporized within solid tumors and the resultant bubbles driven acoustically to enhance HIFU-mediated heating.

Journal of
The long-term goal is to use PSNE to decrease the acoustic pressure and treatment time required to ablate solid tumors, in order to improve the clinical feasibility and safety of MR-guided HIFU for non-invasive treatment of non-resectable solid tumors.Previous studies have reported adverse events that were observed with HIFU ablation treatments, in particular skin burns [31,32].It is anticipated that skin burns (and potentially other adverse events) will be reduced by using PSNE to lower the acoustic pressures required to ablate tumors, thereby increasing the safety of this procedure.

120
Accumulation of Phase-Shift Nanoemulsions to Enhance MR-Guided Ultrasound-Mediated Tumor Ablation In Vivo Future studies will seek to improve detection and activation of PSNE for ultrasoundmediated ablation of solid tumors, but the results of the current pilot in vivo study suggest that PSNE may enhance the efficiency of MR-guided HIFU for treatment of non-resectable solid tumors.

CONCLUSIONS
It was found with T 1 -weighted MRI that gadolinium-tagged PSNE accumulated in rabbit VX2 tumors for at least six hours after intravenous injection.B-mode ultrasound imaging revealed that PSNE were vaporized into microbubbles in vivo using short HIFU pulses.Furthermore, MR thermometry measurements showed that PSNE enhanced HIFU-mediated heating in rabbit tumors.These results suggest that PSNE may potentially improve the clinical feasibility of MR-guided HIFU by reducing the treatment times and acoustic intensities required.
Figure 1.Illustration of a perfluorocarbon droplet in PSNE.A droplet is composed of a liquid perfluorocarbon core surrounded by a phospholipid shell containing polyethylene glycol (PEG) to improve stability in vivo.

Figure 3 .
Figure 3.Size distribution of Gd-PSNE and standard PSNE, measured with dynamic light scattering.The units of the ordinate axes are based on the intensity of scattered light from particles of a certain size relative to the total scattered light intensity from the sample.

116
Accumulation of Phase-Shift Nanoemulsions to Enhance MR-Guided Ultrasound-Mediated Tumor Ablation In Vivo Before PSNE injection 0.5 hrs after PSNE injection 2 hrs after PSNE injection 6 hrs after PSNE injection

Figure 4 .
Figure 4. Representative T 1 -weighted MR images of Gd-PSNE accumulation in tumor as a function of time after injection.Signal enhancement from Gd-PSNE accumulation is evident throughout the periphery of the tumor (indicated with red circles).

)Figure 5 .Figure 6 .
Figure 5.Total tumor volume with enhanced MR signal from Gd-PSNE as a function of time after injection.Error bars represent the standard deviation from four rabbits.

Figure 7 .Figure 8 .
Figure 7. Relative difference in temperature due to HIFU-mediated heating with and without PSNE, plotted as a function of time during and after sonication.The gray box indicates the time when the tumor was sonicated with an acoustic power of 5.2 W. Error bars represent the standard deviation of the data from three rabbits.